FastKernels

LLMs and new frameworks boost GPU kernel optimization

Researchers are exploring novel ways to optimize GPU kernel performance for large language models. One approach uses language models as surrogates to predict kernel performance, significantly increasing the number of ca…